Benchmark menu

Model detail

GPT-OSS 120B

A static, shareable model page that remains complete without JavaScript.

GPT-OSS 120B

Provider: OpenAI · Public slug: openai-gpt-oss-120b

Rank
30
Rank movement
Baseline
Score
23.5
Last measured
2026-07-05
Coverage
None not supported

Reasoning scores

XHigh
27.7
Medium
19.3
None

Generation health

XHigh0%63%37%n 8Medium13%75%12%n 8 None not supportedHigh repair rateHigh failure rate

Cost summary

Mean standard list-price equivalent
$0.0056
Median unique-cell cost
$0.0034
Unique-cell range
$0.0011–$0.02
Coverage
16 unique cells / 16 observed runs
Observed-run total
$0.09
Evidence basis
Provider-reported
Mean output
17k tok
Output range
2.0k–44k tok
Token samples
16

Cost means equal-weight unique model/game/reasoning cells. Provider-account billed spend is reported separately on Charts.

Run setup

  • Overall Models: combines public reasoning variants
  • Standard GameBench code-generation prompt